我目前正在学习3D计算机图形学，并将并行投影规范化为canocial view volume（LookAt Matrix作为熟悉的名字）。我尝试使用纯javascript作为下面的参数将其实现到代码中。var VRP = new Vertex(0,0,0);var VPN = new Vertex(0,0,1);var VUP = new Vertex(0,1,0);var PRP = new Vertex(8,8,100);var Window = [-1,17,-1,17];var F = 1, B = -1;现在，这是我的尝试。我首先将其转换为正交视图体积。注意：您可以将这些步骤直接跳到此处的代码中，并帮助我修复代码以将立方体向前移动到相机（屏幕）而不是移开1. 将 VRP 转换为源var TVRP = [];TVRP[0] = [1, 0, 0, -VRP.x];TVRP[1] = [0, 1, 0, -VRP.y];TVRP[2] = [0, 0, 1, -VRP.z];TVRP[3] = [0, 0, 0, 1];2. 旋转 VRC，使 n 轴、u 轴和 v 轴按顺序与 z 轴、x 轴和 y 轴对齐function normalizeViewPlane(VPN) { var unitVector = calculateUnitVector(VPN); //VPN/|VPN| return normalizeVector(VPN,unitVector);}function normalizeViewUp(VUP, n) { var dtProd = dotProduct(n,VUP); var nVUP = new Vertex(n.x*dtProd, n.y*dtProd, n.z*dtProd); VUP = new Vertex(VUP.x-nVUP.x, VUP.y-nVUP.y, VUP.z-nVUP.z); var unitVector = calculateUnitVector(VUP); //VUP/|VUP| return normalizeVector(VUP,unitVector);}function normalizeUVN(n,u) { var V = crossProduct(n,u); var unitVector = calculateUnitVector(V); //V/|V| return normalizeVector(V,unitVector);}var n = normalizeViewPlane(VPN);var v = normalizeViewUp(VUP, n);var u = normalizeUVN(v, n);var RVRC = [];RVRC[0] = [u.x, u.y, u.z, 0];RVRC[1] = [v.x, v.y, v.z, 0];RVRC[2] = [n.x, n.y, n.z, 0];RVRC[3] = [0, 0, 0, 1];//Perform matrix multiplication 4x4 R.T(-VRP)var res = multiplyMatrix4x4(RVRC, TVRP);3. 剪切 DOP 变得平行于 z 轴function shearDOP(PRP, uMaxMin, vMaxMin) { var CW = new Vertex(uMaxMin,vMaxMin,0); var mPRP = new Vertex(PRP.x,PRP.y,PRP.z); return new Vertex(CW.x - mPRP.x, CW.y - mPRP.y, CW.z - mPRP.z);}var uMaxMin = (Window[1]+Window[0])/2;var vMaxMin = (Window[3]+Window[2])/2;var DOP = shearDOP(PRP,uMaxMin,vMaxMin); var HX = (DOP.x/DOP.z)*-1;var HY = (DOP.y/DOP.z)*-1; var Hpar = []; Hpar[0] = [1,0,HX,0]; Hpar[1] = [0,1,HY,0]; Hpar[2] = [0,0,1,0]; Hpar[3] = [0,0,0,1]; //res = R.T(-VRP) res = multiplyMatrix4x4(Hpar,res);将其转换为类视图体积后，我决定将立方体顶点乘以此最终结果转换矩阵。
添加回答
举报
0/150
提交
取消